An ad that appeared in thousands of Facebook feeds this summer featured an altered video of the White House press secretary, Karoline Leavitt, at a regular press briefing. In it, she appeared to say Americans could claim a $5,000 relief check on an official government site. An arrow that then appeared instead led to an advertiser called Get Covered Today.

Similar ads showed fabricated videos of Senators Bernie Sanders of Vermont and Elizabeth Warren of Massachusetts promising similar rebates that did not exist. “This is not a gimmick,” the impersonation of Ms. Warren says.

In fact, it was.

Even so, the company behind the ads and others like it were among the top political advertisers on Facebook, according to an analysis by the Tech Transparency Project, a nonprofit focused on holding large technology companies accountable.

The ads are a lucrative part of Facebook’s advertising revenue that, the project’s researchers and others say, has led the company to turn a blind eye to a flood of low-quality or deceptive content, spam and in some cases outright fraud on the platform.

“Meta is very aware of these types of scams,” said Katie A. Paul, director of the Tech Transparency Project. “They just didn’t care.”

In a report released on Wednesday, the project identified 63 advertisers that have by several measures employed deceptive or fraudulent practices. They amount to roughly one in five of the platform’s top 300 spenders in the category of political or social advertising.

Collectively they bought nearly 150,000 ads, spending almost $49 million over the past seven years, according to data in Facebook’s ad library.

Ads from all of the 63 advertisers have previously been removed for violating Facebook’s policies, meaning their deceptive practices were not unknown. Meta has suspended some of them, but more than half were able to continue posting new ads as recently as this week. The analysis suggests that even when Facebook takes action, it has profited from advertisers that try to swindle its users.

“The only thing worse than these deepfake A.I. videos being used to scam Americans is the fact that Meta makes tens of millions of dollars off those scam ads,” Senator Warren said in a statement about the impersonation of her.

She added, “We need serious guardrails in place to protect consumers online.”

Facebook, which is owned by Meta, prohibits advertising that uses “identified deceptive or misleading practices, including scams to take money from people or access personal information.” It has explicit rules against impersonations, and it ultimately removed the video of Ms. Leavitt, though not all the others.

In a statement, the company said it enforced its rules vigorously and would “invest in building new technical defenses” against what it called an industrywide issue. “Scammers are relentless,” the statement said, “and constantly evolve their tactics to try to evade detection.”

Facebook has long had to wrestle with criticism of the political ads it solicits on the platform. It imposed restrictions on them after Russia used them in an attempt to sow voter discontent during the 2016 presidential election and banned them entirely for a time after President Trump fought to overturn the 2020 election.

With government officials around the world reporting a sharp rise in online fraud, Meta is facing new pressure.

Last week, Singapore’s government gave Meta until the end of the month to crack down on scam ads and other posts after a sharp rise in content impersonating officials there. The government threatened a fine starting at $770,000 and rising daily if Meta did not comply.

In the United States, the company argued in court last year that it “does not owe a duty to users” to address fraudulent content, but that legal argument appears to be falling short.

A federal court in California refused last week to dismiss a lawsuit that accused Facebook of negligence and breach of contract for abetting fraud by advertisers.

“While Facebook disclaims responsibility for third-party conduct, it maintains responsibility for its own promise to ‘take appropriate action’ to combat scam advertisements and to do so in good faith,” Judge Jeffrey S. White of U.S. District Court for the Northern District of California wrote in his ruling.

Ad revenue on Meta’s platforms, which also include Instagram and WhatsApp, reached $160 billion last year. The analysis of political ads on Facebook represents only a small fraction of the total.

Meta keeps political ads in its library for seven years as a measure of transparency after the Russian influence operation in 2016. Ads in other categories are not preserved after they appear, making it harder for researchers to study the extent of scams among those.

Facebook has specific requirements for political advertisers, requiring them, for example, to be registered companies in the United States. The process of submitting and approving ads that appear on the platform is largely automated, allowing dishonest actors to slip through, experts say.

Some of the advertisers studied by the Tech Transparency Project included contact information in the Philippines, Bangladesh, Vietnam or Britain, which should presumably have disqualified them.

“You can be anybody,” said Iesha White, director of intelligence for Check My Ads, an advocacy group that monitors the digital ad industry and was not involved in the Tech Transparency Project’s analysis. “There are no rigorous checks.”

Meta, she said, effectively crowdsources its enforcement, waiting for users or researchers to report fraud. “It should be identified before the ads ever run.”

Many of the ads appear to target older and conservative users. They often respond quickly to politically charged events.

On the day of the memorial service for Charlie Kirk on Sept. 21, an ad appeared asking if people could forgive his assassin, as his wife, Erika, said she had. Click to vote now, it said, promising a red “We Are Charlie Kirk” hat for those who did — free, except for 99 cents to help cover the shipping costs.

Though the hat was ostensibly free, it required a more costly monthly membership for a clothing and other merchandise supplier calling itself End the Wokeness. Entering a credit card to pay for the nominal shipping meant automatically signing up for the subscription.

The Federal Trade Commission has warned consumers about websites that sign them up for subscriptions they do not want.

The company that operates as End the Wokeness, Sculpin Media, has listed page administrators in the Philippines, as well as the United States. It describes itself as a clothing store and media company and operates out of a small office next to a spice dealer and a car detailer in a low-slung brick warehouse near the former Fresh Kills landfill in Staten Island.

A man who identified himself as a co-owner of the company appeared surprised that a reporter had found the office. He acknowledged placing the ads but declined to speak on the record. He said the staff worried about potential threats of political violence.

Inside the office were boxes of red baseball caps with the slogan MAGA 2028. End the Wokeness’s website is devoted to sales of Trump- and MAGA-themed knickknacks. Among the latest is a “Patriot Pumpkin Carving Kit,” a stencil that features Mr. Trump’s mug shot. The company’s own page on Facebook includes numerous posts from users complaining about its subscription model or asking for refunds.

In August, the Federal Trade Commission warned that reported scams targeting older Americans by impersonating government officials or businesses had increased fourfold between 2020 and last year.

Frauds on social media platforms and other websites now far exceed those using text messages or phone calls. The spread of commercially available tools using artificial intelligence has clearly contributed to the surge in deceptive ads, making them easier to produce and spread.

“This is part of a much larger issue: A.I. is advancing rapidly, but the laws and protections we have in place for everyday Americans are dangerously outdated,” Mr. Sanders said in a statement in response to an inquiry about the deepfake of him.

The ad featuring Ms. Leavitt was posted by one of four advertisers linked to a company calling itself RFY News Group. No one responded to a message left at a number listed in the ad library as a contact.

Neither Ms. Leavitt nor the White House responded to a request for comment about the deceptive videos.

Meta suspended the accounts of two of the four advertisers involved in August, though they had been advertising for weeks at that point. Two others appeared to have been removed this week following inquiries about the Tech Transparency Project’s research, which has tracked the RFY News Group for more than a year.

Researchers say Meta could do more to limit ads like these and to be more transparent about its enforcement of its own rules. Instead, it has rolled back its efforts to moderate political content and reduced its teams working on platform safety.

“The pendulum has swung toward more short-term revenue considerations versus the long-term health of the platform and the ecosystem,” said Rob Leathern, a former Meta employee who helped create the ad library and is now head of Hawkview Labs, a company that helps start-ups working in online trust and safety. “Perhaps at some point it has to swing the other way.”

Eli Tan in San Francisco and Mark Bonamo in New York contributed reporting.